Guaranteed deadlines with dynamic recovery blocks in distributed systems

In a hard real-time system, a timing fault is said to occur when a real time task delivers its results too late. An deadline mechanism has been proposed to avoid timing faults in uniprocessor machines. Under this approach, every task is implemented with two algorithms: on one hand, a primary program produces a good quality service but in an unknown and possibly infinite length of time due to software errors; and on the other hand, a back-up alternate program produces an acceptable result in a known and fixed length of time. We propose an extension of this approach to distributed systems. We present a task allocation strategy to cope both with processor failures and timing faults